Predictive Maintenance

Predictive maintenance is the practice of using data to try to anticipate and avoid machine failures. It differs from preventative maintenance in that it uses checks to identify abnormal conditions before the machine has a problem.

Most failures in machines are not immediate, catastrophic failures. Electric motors might show changes in the amperage they draw, for example, before they finally stop working. Vibration might be an indicator of a problem with a cutting tool. If you can identify the link between those indicators and the failure, you can set up a predictive maintenance program.

A good example of predictive maintenance is the Army Oil Analysis Program. In this program, oil samples from vehicles are sent to a lab on a regular schedule. The lab checks for a variety of indicators. It can check for things like metal shavings, contaminants, and degradation of the lubricant. The results are compared to historical data and are used to identify impending failures.

So, imagine a component started to fail, and released metal shavings into the oil. Preventative maintenance would check for oil leaks, an indicator that seals may have been damaged by the shavings. You still have to replace the failed component, but the seals are now also bad, and there is a chance that the component has become damaged beyond repair. Worse, the shavings could also have damaged additional components, amplifying the problem.

Predictive maintenance, on the other hand, would detect minute shavings in the lubricant before they build up and cause damage. The component could be replaced or repaired, and the seals would still be intact, preventing additional repairs.

Creating a Predictive Maintenance Program

Now, in practice, it can be a challenge to find good predictive maintenance checks. This is hard for manufacturers, but even more so for end users, who see a lot fewer failures. But be creative and keep looking.

You may be able to isolate instrument readings that predict failure. Perhaps the current an electric motor is drawing is rising slightly over time. Maybe the machine has excessive vibration. Perhaps the life of a tool changes from what you would expect based on usage. The fuel consumption of a gas motor could change. A machine may develop hotspots.

All of these things can be indicators of pending failures.

Preventative Maintenance vs Predictive Maintenance

It is easy to confuse predictive and preventive maintenance. For example, if oil is low, it is generally an indicator of a problem. You could think of it as a simple maintenance check—after all, there is a problem if you are losing oil. You could think of it as preventive. Adding oil will likely keep a bigger problem from occurring. Or it could be predictive. It could be linked to a specific failure that you could see coming in the future.

In practice an operator or a maintenance technician doesn’t really care how you categorize the check. They know that they are supposed to check the oil before use, and if it is low, a designated person takes a prescribed action.

Where the distinction between preventive and predictive does matter, though, is when you are setting up a maintenance program. The best checks are predictive. They let you know in advance when a problem is coming. They can save time and money because you only spend repair resources right before a machine needs it. You are not doing work when there is still useful life left in parts. And you are not replacing parts or fluids more frequently than needed, which limits workload on the maintenance team and downtime on machines.

Knowing to look for predictive checks is an important component of an effective maintenance program.